Club of Amsterdam, Future, Think Tank ..

Club of Amsterdam Journal
Free Subscription
click here

























































































































keyword search

Training Architectural Computational Critics by Example
Average reader rating: 0  
by Reid E. Williams 14 the future of Architecture

New building technologies and materials coupled with a modular construction system o er consumers an unprecedented chance to customize their living spaces. At the center of this customization process is a computa- tional tool that guides consumers through the process of designing a home or apartment. Algorithms for architec- tural computational critics that are trained by a designer through examples and that can then critique designs is proposed as part of the design tool. A prototype system encompassing two apartment design scenarios is built and tested. The prototype demonstrates the ability to learn architectural concepts through training.

The face of urban and suburban housing is poised to be changed by a wealth of new computational, material, and manufacturing technologies. The design, construc- tion, and industry of housing are all likely to be e ected by this technology. Increas- ingly sophisticated materials and manufacturing techniques will make a modular, component-based construction system both feasible and a ordable [11]. Components will be manufactured o -site and then assembled on-site. This construction system enables two important properties of future homes and apartments. First, in apartment buildings, the construction of individual apartments can be decoupled from that of the apartment building shell, allowing for the construction of the building to begin (and even nish) before individual apartments themselves are designed and built. Second, since manufacturing of components takes place o -site, on-site construction becomes on-site assembly, thereby reducing on-site construction time and cost. These two features are the rst step towards mass customizable housing: giving people who would otherwise live in \cookie-cutter" apartments or subdivisions the ability to cre- ate a custom living environment that is architecturally sensible and that ts their lifestyle.

To fully realize the power of such a system, non-experts must be given the power to design their own living spaces. Direct, traditional interaction with an architect is prohibitively expensive for all but the wealthiest consumers. An alternative is a computer based design tool with which a non-expert interacts to design a suitable living space. This tool will allow consumers to economically design and customize their living spaces. This work proposes a means to create a tool that provides typical homeowners with architectural expertise via computerized architectural \critics". To be useful, the tool must be easy to use and easy to extend with new architectural styles or viewpoints. The system described in this work allows architects to train new architectural computational critics by example. The system is implemented in a two dimensional oor-plan scenario, although its infrastructure is suitable for more complex design tasks.


The work presented proposes an infrastructure for creating computational critics that are trained by example to evaluate residential designs. Prior work has focused on creating design tools that are either not intended for non-experts or are intended for non-experts but require a signi cant amount of knowledge engineering. The critic infrastructure proposed here interfaces with a design tool that is intended for non- experts and requires little explicit knowledge engineering e ort. Results show that the proposed infrastructure is feasible and promising as a means to provide architectural feedback in a non-expert design tool.

You can download the full report: click here

Visit also the conference about the future of Architecture and the sections with books, articles and links.

Rated 0 by other users. What do you think? [rate this article]

Copyright 2002-2020 Club of Amsterdam. All rights reserved.    Contact     Privacy statement    Cancellation Policy